Batch Picking Clustering Model for Order Fulfillment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional in-store picking procedures face challenges in efficiently completing pick-walks due to limitations in accurately determining travel distances between items, which affects the optimization of pick-walk routes and the overall efficiency of order fulfillment.
Innovation Solution
The method involves generating and optimizing batch picks by using a clustering model that utilizes vectorized representations of items, including location information such as walking distances in time between zones, categories, and subcategories, to assign close items together in a pick-walk and accurately estimate travel times.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional batch picking procedures are used with fixed limitations on maximum orders per pick-walk, then order fulfillment can be completed with simple routing, but the total number of pick-walks increases and time consumption increases
Solution Approach 1:
The system segments the batch of orders into multiple pick-walks based on item clustering and spatial proximity. By dividing the fulfillment task into smaller, geographically grouped pick-walks, the system reduces travel distance within each pick-walk while maintaining overall productivity through parallel execution of multiple pick-walks.
Solution Approach 2:
The system transitions from traditional single-dimension batch processing to multi-dimensional optimization by considering spatial location, item proximity, and pick-walk routing simultaneously. This enables the system to minimize travel time within each pick-walk while maximizing the number of orders fulfilled per period through optimized clustering.
2Loss of time
If pick-walk routes are not optimized based on accurate travel distance data, then routing complexity is reduced, but the time taken for actual picking increases
Solution Approach 1:
The system performs preliminary actions by pre-calculating travel distances between all item locations and storing them in a distance matrix before batch picking begins. This pre-computation enables rapid route optimization during actual pick-walk execution without adding real-time complexity to the picking process.
Solution Approach 2:
The system creates a virtual representation of the store layout and item locations through a distance matrix that copies spatial relationships into a computable format. This digital model allows for efficient routing optimization algorithms to determine optimal pick-walk paths without requiring complex real-time spatial calculations.
3Device complexity
If items are not clustered based on spatial proximity, then batch picking can be simplified, but the number of pick-walks required increases
Solution Approach 1:
The system changes the parameter of item grouping from traditional category-based or order-based batching to spatial proximity-based clustering. By using the distance matrix to group items that are physically close together, the system reduces the number of pick-walks needed while keeping the clustering algorithm computationally efficient through parameter optimization.
Data Source
AI summary
A system and method for generating pick-walk data from batch picking is provided. The method includes generating a vector for each item in a store, with the vector being generated based on walking distances in time from a zone corresponding to the item to other zones in the facility, walking distances in time from the item to other items within the zone corresponding to the item, a category corresponding to the item, and a subcategory corresponding to the item. Items within an online order are partitioned into clusters with each cluster including items in close proximity to one another. Pick-walks are assigned to a plurality of pickers based on the clusters, witch each pick-walk including items from a plurality of different orders.